Measuring Self Control Problems

American Economic Association

Measuring Self-Control ProblemsAuthor(s): John Ameriks, Andrew Caplin, John Leahy, Tom TylerSource: The American Economic Review, Vol. 97, No. 3 (Jun., 2007), pp. 966-972Published by: American Economic AssociationStable URL: http://www.jstor.org/stable/30035029Accessed: 09/07/2010 11:58

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/action/showPublisher?publisherCode=aea.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to TheAmerican Economic Review.

http://www.jstor.org

Measuring Self-Control Problems

By JOHN AMERIKS, ANDREW CAPLIN, JOHN LEAHY, AND TOM TYLER*

While models of self-control problems have proliferated in recent years, there have been few corresponding advances in measurement. We develop a survey instrument to measure self- control problems and apply it to a sample of highly educated adults. Measured self-control relates in the anticipated manner to wealth accumulation and standard personality measures. Yet while self-control problems are typically seen as resulting in overconsumption and low wealth, we identify a significant group who underconsume and thereby accumulate high levels of wealth. In addition, self-control problems are found to be smaller in scale for older than for younger respondents. Those who put money aside in retirement accounts may be delaying access to a point at which self-control problems are no longer important. Continued advances in measurement are essential to guide development of self-control models in empiri- cally relevant directions.

I. The Self-Control Measure and Its Properties

Most theories of self-control share a common structure. There is an ideal action that the agent

* Ameriks: Vanguard, P.O. Box 2600, MSV37, Valley Forge, PA 19482 (e-mail: [email protected]); Caplin: Department of Economics, New York University, 19 West 4th St., New York, NY 10012 (e-mail: [email protected]); Leahy: Department of Econom- ics, NYU, 19 West 4th St., New York, NY 10012 (e-mail: [email protected]); Tyler: Department of Psychology, NYU, 6 Washington Place, New York, NY 10003 (e-mail: [email protected]). Caplin thanks the Center for Experi- mental Social Science at New York University and the C.V. Starr Center at NYU for financial support. We would like to thank Douglas Bernheim, Xiaohong Chen, Douglas Fore, Faruk Gul, Guido Imbens, David Laibson, Brent Roberts, Ariel Rubinstein, Julio Rotemberg, Yaacov Trope, and three anonymous referees for their help. We gratefully acknowl- edge financial support for our survey provided by the TIAA-CREF Institute and to Caplin and Leahy under National Science Foundation grant SES-0351115. The research described in this article was completed while Ameriks was Senior Research Fellow at the TIAA-CREF Institute. The opinions expressed in this article are solely those of the authors, and do not necessarily reflect the views of their current or past employers.

would like to take and there is something that tempts the agent to deviate from this ideal. The actual action represents a balance between these forces. Models that fit this general framework include the model of temptation and self-control of Faruk Gul and Wolfgang Pesendorfer (2001); the time-inconsistent framework of Robert Strotz (1956) and David Laibson (1997); the model of cue-triggered mistakes of B. Douglas Bernheim and Antonio Rangel (2004); and the dual-self models of Richard Thaler and Hersch Shefrin (1981), Jess Benhabib and Alberto Bi- sin (2005), and Drew Fudenberg and David Levine (2004).

A. The Question Our measure of self-control problems makes

use of this structure. It is based on a simple hypothetical choice scenario. We assume that people understand whether they face a control problem and know how it affects their choices. We ask people how they would ideally allocate a prize over time, whether they would be tempted to deviate from this ideal, and whether their actual choice would deviate from the ideal. To ensure the allocational integrity of our hypothetical problem, we bound the period of availability of the prize. To remove simple sub- stitution into the general lifetime pattern of consumption, we want the prize to be attractive, yet too much of a luxury for most agents to pay for out of their own resources. We also do not want the prize to be a completely indivisible, once- in-a-lifetime experience, since this would re- duce the information content of our allocation question. Extraordinary restaurant meals struck us as a good candidate of close-to-universal appeal. We asked the following question:

Suppose you win ten certificates, each of which can be used (once) to receive a "dream restaurant night." On each such night, you and a companion will get the best table and an unlim- ited budget for food and drink at a restaurant of your choosing. There will be no cost to you: all payments, including gratuities, come as part of the prize. The certificates are available for im-

966

VOL. 97 NO. 3 AMERIKS ET AL.: MEASURING SELF-CONTROL PROBLEMS 967

mediate use, starting tonight, and there is an absolute guarantee that they will be honored by any restaurant you select if they are used within a two-year window. If they are not used up within this two-year period, however, any that remain are valueless.

The questions below concern how many of the certificates you would ideally like to use in each year, how tempted you would be to depart from this ideal, and what you expect you would do in practice:

(a) From your current perspective, how many of the ten certificates would you ideally like to use in year 1 as opposed to year 20

(b) Some people might be tempted to depart from their ideal allocation in (a). Which of the following best describes you (please mark only one): - I would be strongly/somewhat tempted

to keep more certificates for use in the second year than would be ideal;

- I would have no temptation in either direction (skip to d);

- I would be somewhat/strongly tempted to use more certificates in the first year than would be ideal.

(c) If you were to give in to your temptation, how many certificates do you think you would use in year 1 as opposed to year 20

(d) Based on your most accurate forecast of how you think you would actually behave, how many of the nights would you end up using in year I as opposed to year 20

Our measure of self-control problems is the numerical difference between expected consumption in the first period and ideal consumption, (d) less (a). We label this difference the expected-ideal (EI) gap. A positive El gap represents a standard problem of overconsumption, while a negative gap corresponds to underconsumption. We can also construct a measure of temptation, results on which are reported in Section III below.

B. The Sample

Our questions were included in a survey sent in February 2003 to a sample of TIAA-CREF participants. All of the approximately 2,500 who received the survey had responded to two previous surveys: the Survey of Participant Fi-

nances (henceforth SPF), fielded in January 2000; and the Survey of Financial Attitudes and Behavior (henceforth FAB), fielded in January 2001. The response rate to our third survey was on the order of 65 percent, with 1,632 providing responses. We removed 87 respondents who failed to answer both the questions on expected and ideal consumption. We also asked respondents to value the free dinner prize and removed 25 respondents for whom the prize had no value. We end up with 1,520 in the "entire sample," which defines the sample in analyses that do not require complete data on wealth.

In analyzing wealth accumulation, we limit attention to the subsample that supplied complete data on all financial and demographic variables of interest. The asset and debt information is drawn from the SPF, in which a highly detailed breakdown of wealth by category is available. (The results in Ameriks, Caplin, and Leahy (forthcoming) comparing self reports with accounting data, indicate the wealth data to be of unusually high quality.) Data on earnings are from the FAB, in which we asked households to provide esti- mates of their overall taxable income from employment in 1999, as well as past and projected future income from employment. We eliminate a total of 1,015 respondents due to incompleteness of data, primarily in the wealth and income categories. We also drop 128 annuitants for whom data on retirement assets are hard to interpret, and 3 outliers with unusually large gross financial assets in excess of $5 million (inclusion of these additional 131 subjects leaves the results essen- tially unchanged). We refer to the 374 remaining households as the "regression sample."

Our working paper (Ameriks et al. 2004) tabulates the basic demographic, educational, occupational, and economic characteristics of households, in both the entire sample and the regression sample. As detailed therein, our sample is far from representative. Respondents stand out in terms of their educational achieve- ments and their financial status. Just over one- third of respondents have PhD degrees. Median net worth (gross financial assets and real estate assets less total debt) is about $500,000, far higher than among working households in the 1998 Survey of Consumer Finances. The vast majority of households have significant nonretirement financial assets, and very few have high levels of personal debt. The median level of personal debt is zero.

968 THE AMERICAN ECONOMIC REVIEW JUNE 2007

TABLE 1-DISTRIBUTION OF THE El GAP

Entire sample Regression sample EI gap Number Percent Number Percent

5 9 0.6 2 0.5 4 2 0.1 0 0.0 3 8 0.5 1 0.3 2 39 2.6 9 2.4 1 113 7.4 35 9.4 0 1,059 69.7 246 65.8

-1 141 9.3 41 11.0 -2 94 6.2 28 7.5 -3 25 1.6 7 1.9 -4 9 0.6 1 0.3 -5 14 0.9 2 0.5 -6 2 0.1 1 0.3 -7 1 0.1 1 0.3 -8 1 0.1 0 0.0 -9 3 0.2 0 0.0 All 1,520 100.0 374 100.0

Source: Authors' tabulation of 2003 survey data.

C. Ideals, Expectations, and Corners

Nearly 60 percent of respondents indicated that their ideal allocation involved an equal split between the two periods. Among those who gave other answers, the overwhelming tendency was to wish to consume more in the first year, with almost eight times as many selecting answers of six and above than answers of four and below. The contrast at the extremes is especially striking. More than 15 percent of respondents stated a wish to consume all of their meals in the first year, with only a tiny fraction preferring to consume all in the second year. The distribution of expected consumption is more dispersed, with less than 50 percent expecting an equal split.

Table 1 reports the distribution of the El gap for both the entire sample and the regression sample. The El gap is typically small: 95 percent of responses are less than two in absolute value. Roughly two in every three respondents have El gaps of zero, corresponding to their having no self-control problem according to our measure. Note, also, that of those with a nonzero El gap and therefore a measured problem of self-control, roughly two in every three expect to use fewer than their ideal number of certificates in the first year, with only one in three expecting to use more than their ideal number. This suggests that there is a significant group who appear to have problems of under-

consumption, at least for consumption activities that also involve time.

Either the expected or the ideal consumption lies at a comer for about 17 percent of the observations. It is possible that two individuals may have identical self-control problems yet different measured El gaps, if differences in their ideal levels of consumption lead one or both to hit a corner. Our measure of self-control problems is therefore censored. We address this issue in our statistical analysis.

II. Self-Control Problems and Wealth

We investigate the relationship between self- control problems and wealth in a regression of the form

(1) w = a + SC + 2X + ,

where w is some relevant wealth measure, sc is the self-control problem, x is a vector contain- ing other economic and demographic variables often included in classical life-cycle regressions, and e is an error term. We use 1999 income from the FAB as our right-hand-side income variable, since this corresponds most closely to the wealth data from the SPF.

Before running the regression, we outline an imputation procedure designed to resolve the censoring problem. We know that the right- censored observations are greater than or equal to the El gap and the left-censored observations are less than or equal to the El gap. We therefore first estimate f(sclx) from the regression,

El gap =/30 + 3x + v.

Next, we replace the censored observations with draws from f(sclx, sc

-

El gap) orf(sclx, sc -

El gap), depending on the direction of the censoring. We repeat this imputation procedure ten times and take as our estimate of a1 the average of the estimated

&l's. Table 2 summarizes overall regression results when the wealth variable is total net worth (nonretirement financial assets plus retirement financial assets, plus real estate assets, less total debt). The regression identifies a clear relationship between self-control problems and wealth accumulation. Note that we also include the answer to


TABLE 2-NET WORTH REGRESSION RESULTS

Variable Coeff. Std. err.

Expected-ideal gap -0.146*** 0.048 Ideal level -0.019 0.033 Log 1999 income 0.198 0.179 Zero 1999 income 1.555** 0.776 Past income 0.469*** 0.161 Zero past income 1.304* 0.707 Future income -0.047 0.109 Zero future income -0.190 0.467 Age 0.216*** 0.046 Age2 -0.001*** 0.000 Empl. status

Working Omitted Partially retired 0.068 0.224 Retired 0.267 0.264

Occupation Faculty Omitted Mgmt./sen. admin. -0.185 0.155 Tech./professional 0.003 0.147 Other -0.134 0.174

Education College or below -0.236 0.172 M.A./professional Omitted Ph.D. 0.051 0.128

R. has defined ben. plan -0.222* 0.127 S. has defined ben. plan -0.087 0.157 Marital status

Curr. married Omitted Prev. married -0.601*** 0.169 Never married -0.345** 0.158

Male respondent -0.061 0.113 Num. kids 0.013 0.063 Constant -3.356*** 1.127 N 374

Notes: The dependent variable is log of net worth. We used a censored regression (Tobit) technique to include (3) people with net worth of zero or less. Log income was set to zero for those with zero income. Asterisks indicate the level of statistical confidence for rejection of the hypothesis that the relevant coefficient is (independently) equal to zero: *** indicates rejection at better than a 1 percent level of confidence, ** indicates rejection at better than a 5 percent level, and * indicates rejection at better than a 10 percent level. The Pseudo-R2 was 0.2417. Source: Authors' tabulation of 2003 survey data.

question (a) on the ideal level of consumption and find it to have no explanatory power whatsoever. In quantitative terms, the equation suggests that the average overconsumer accumulates some 20 percent less than one with no self-control problem, while the average underconsumer accumulates some 25 percent more.

The finding of a significant impact of self- control problems on net worth is robust to alternative treatments of the corner constraints. Since we get almost identical results when we

ignore the corner constraints, we will ignore this issue in the remainder of the paper. The finding is also robust to the removal of regressors from the right-hand side, and to the introduction of additional regressors, such as measured preference parameters, information on parental gifts and bequests, and wealth shocks. Restricting the sample to those under 65 shrinks the sample to 326, yet increases the absolute value of the coefficient on self-control problems, as well as its statistical significance. Adding annuitants lowers the parameter somewhat, but significance remains.

Most theories of self control suggest that there is a significant difference between liquid and illiquid assets: it is harder to resist the temptation to consume out of liquid assets. Ta- ble 3 shows that our measure of self-control problems does indeed appear to have a greater impact on liquid assets than on illiquid assets. The liquid assets we analyze are nonretirement financial assets. The less liquid assets are retirement financial assets (note that real estate assets and debt are not included in either regression). The sample for these regressions is restricted to the group age 64 and under, since the asymmetry in liquidity between retirement and nonretirement assets is reduced when individuals reach the age of retirement. The relationship between measured self-control problems and nonretirement financial assets is robust to all variations in the treatment of the corner constraints and to the addition and removal of regressors.

III. The El Gap as a Measure of Self-Control

A. Psychological and Demographic Correlates

Personality psychologists associate self-control with conscientiousness, one of the "big five" personality factors. If the El gap is a measure of self-control, we would expect it to be correlated with measures of conscientiousness. We asked respondents to evaluate themselves on a six-point scale of agreement and disagreement using two standard conscientiousness questions from Paul T. Costa and Thomas A. Widiger (1994): "Sometimes I am not as dependable or reliable as I should be"; and "I never seem able to get organized."

Table 4 reports the results of a regression of the EI gap on age, sex, and our two measures of


TABLE 3-REGRESSIONS FOR WEALTH CATEGORIES

Non-ret. fin. assets Retirement assets

Variable Coeff. S.E. Coeff. S.E.

Actual-ideal gap -0.285*** 0.079 -0.081 0.055 Ideal level -0.006 0.057 0.018 0.040 Log 1999 income 0.059 0.306 0.091 0.216 Zero 1999 income 1.336 1.601 1.492 1.130 Past income 0.856*** 0.300 0.540** 0.211 Zero past income 3.297* 1.743 1.272 1.230 Future income -0.033 0.181 -0.054 0.128 Zero future income 0.366 0.797 -0.119 0.562 Age -0.112 0.100 0.281*** 0.071 Age2 0.001 0.001 -0.002*** 0.001 Empl. status

Working Omitted Omitted Partially retired -0.219 0.383 0.430 0.270 Retired -0.299 0.510 -0.038 0.359

Occupation Faculty Omitted Omitted Mgmt./sen. admin. 0.152 0.259 -0.077 0.182 Tech./professional -0.003 0.250 0.076 0.176 Other -0.021 0.300 -0.302 0.211

Education College or below -0.794*** 0.289 -0.264 0.203 M.A./professional Omitted Omitted Ph.D. -0.353 0.219 0.091 0.154

R. has DB plan -0.022 0.222 -0.270* 0.156 S. has DB plan 0.134 0.269 -0.024 0.190 Marital status

Curr. married Omitted Omitted Prev. married -0.207 0.291 -0.544*** 0.205 Never married -0.500* 0.275 -0.347* 0.194

Male respondent -0.144 0.190 0.200 0.134 Num. kids -0.079 0.106 0.000 0.074 Constant 2.296 2.255 -5.595*** 1.591 N 362 362 R2 0.078 0.179

Notes: Dependent variables are natural logarithms of the quantities listed at head of each set of columns. Asterisks indicate the level of statistical confidence for rejection of the hypothesis that the relevant coefficient is (independently) equal to zero: *** indicates rejection at better than a 1 percent level of confidence, ** indicates rejection at better than a 5 percent level, and * indicates rejection at better than a 10 percent level. Source: Authors' calculations based on 2000, 2001, and 2003 survey data.

conscientiousness. The data reveal a strong relationship between the conscientiousness questions and the absolute value of the El gap, and no relationship with the level of the El gap. For those who are conscientious, there is a lower divergence between expected and ideal consumption, regardless of sign.

A particularly interesting finding in Table 4 is the profound reduction in the scale of self- control problems as individuals age, which shows up only when one uses the absolute value of the self-control measure. Older individuals experience fewer self-control problems, either

of overconsumption or underconsumption, than do their younger counterparts. This finding is certainly consistent with the common view that temptation falls with age, and has important connections with actual consumption behavior over the life cycle. Models that allow for such a time-changing self-control parameter retirement may be necessary to explain the absence of a spike in consumption spending at the point when retirement assets become fully liquid.

These results hold if we condition separately on a nonpositive or a nonnegative El gap. Each is separately related to conscientiousness and age.


TABLE 4--CONSCIENTIOUSNESS AND SELF-CONTROL

Variable El gap Absolute El gap

Age 0.003 -0.008*** (0.002) (0.002)

Male 0.048 -0.129** (0.063) (0.056)

Not dependable 0.016 0.070*** (0.029) (0.026)

Not organized 0.057 1.101*** (0.029) (0.026)

Constant -0.306 0.682*** (0.169) (0.150)

N 1489 1489 R2 0.005 0.039

Notes: Asterisks indicate the level of statistical confidence for rejection of the hypothesis that the relevant coefficient is (independently) equal to zero: *** at the 1 percent level and ** at the 5 percent level. Source: Authors' tabulations of 2003 survey data.

B. Temptation and Self-Control

We define the temptation-ideal (TI) gap as the difference between the answers to questions (c) and (a), the most tempting choice and the ideal choice. The correlation between the TI gap and the El gap is about 0.4. The TI gap is also correlated with wealth, but loses all predictive power if the El gap is included in the regression. The TI gap appears to work through the El gap.

Most self-control theories suggest that the El gap should lie somewhere between the TI gap and zero. This is true for 1,173 of the 1,448 respondents for whom we can construct both measures. Among the others, 235 report a TI gap of zero, yet a nonzero El gap. Interestingly, the vast majority of the violations (211) occur when the El gap is negative. It is possible that underconsumers do not fit into the ideal-temptation framework. It may be that temptation lacks meaning for this group (what does it mean to be tempted to consume less0) or they may have trouble consuming at all, possibly because they are busy at work or at home. It is also possible that the El gap is capturing something other than self-control in these cases. When we restrict the sample in the wealth regressions to those that fit the TI framework, that is, those for whom the El gap lies between the TI gap and zero, the effect of the El gap tends to be stronger. The coefficient on the El gap rises in absolute value to -0.19 with a t-statistic of 2.34 on 295 individuals, 56 of whom have nonzero El gaps. For nonretirement financial assets, the coefficient is -0.46 and

the t-statistic is 3.51, whereas for retirement financial assets it is -0.12 with a t-statistic of 1.23. Both these regressions have 329 observations.

C. Commitment and Self-Control

An implication of most theories of self-control is that agents would like to precommit to their desired action. Following our main questions, we asked responders whether they would use an op- tion to restrict some of the certificates for use only in the first year and/or the second year, and if so how many certificates they would like to restrict. We dropped 29 observations due to missing data, 19 observations that restricted more than the al- lotted 10 meals, and 103 whose restrictions made it impossible for them to consume their ideal level. This left 1,369 responses. For this group, we define the signed distance between the expected choice and the constraint set to be the revealed preference (RP) gap, a possible alternative measure of self-control.

In many ways, the RP gap reinforces our earlier findings. The correlation with the El gap is 0.5. Like the El gap, the absolute value of the RP gap is positively related to our measures of conscientiousness and negatively related to age, although the correlation with age is significant only at the 6 percent level. Like the El gap, the RP gap has a large effect on wealth, although again the effect is less statistically significant. The p-value is 7 percent. As with the TI gap, people with overconsumption problems according to the El gap are more likely to have a nonzero RP gap than people with underconsumption problems, indicating again that there might be something different about underconsumption.

In other ways, however, the RP gap presents a different and more complex picture. On the one hand, self-control problems appear less se- vere from the perspective of the RP gap. As was mentioned above, the correlation with wealth is less significant. Surprisingly, there is no correlation between the RP gap and liquid assets. Partly this is because few are willing to impose binding constraints on themselves. Fewer than 10 percent of agents impose strictly binding constraints, while 30 percent have self-control problems according to the EI gap. On the other hand, while binding constraints are rare, non- binding constraints are common. Thirty percent of the respondents with a zero RP gap choose to restrict some certificates to one period or the


other. Over 60 percent of the respondents who restrict certificates to one period also choose to restrict some to the other period.

These findings for the RP gap do not fit our simple theoretical models of self-control problems. The weak relationship with wealth, the unwillingness of some to commit, and the will- ingness of others to overcommit suggest con- siderations other than self-control are at work. Those who do not commit may value the flexibility to adjust their plans more than the cost of missing their target (Manuel Amador, Ivan Werning, and George-Marios Angeletos 2006). Those who overcommit may value the certainty of having definite plans (Ameriks, Caplin, and Leahy 2003). For these reasons, we prefer the El gap as a measure of self-control problems. We cannot, however, rule out the possibility that the El gap is correlated with other factors that strongly affect wealth, and that in other samples a commitment-based measure may be preferable.

IV. Concluding Remarks

We have introduced a survey-based measure of self-control problems that correlates, as theory predicts, with wealth measures, in particular with liquid wealth. While these problems are typically seen as resulting in overconsumption and low wealth, we identify a significant group who underconsume and thereby accumulate high levels of wealth. We also find that self-control problems are smaller for older respondents.

REFERENCES

Amador, Manuel, Ivan Werning, and George- Marios Angeletos. 2006. "Commitment vs. Flexibility." Econometrica, 74(2): 365-96.

Ameriks, John, Andrew Caplin, and John Leahy. 2003. "Wealth Accumulation and the Propen- sity to Plan." Quarterly Journal of Econom- ics, 118(3): 1007-47.

Ameriks, John, Andrew Caplin, and John Leahy. Forthcoming. "Retirement Consumption: In- sights from a Survey." Review of Economics and Statistics.

Ameriks, John, Andrew Caplin, John Leahy, and Tom Tyler. 2004. "Measuring Self-Control." National Bureau of Economic Research Working Paper 10514.

Benhabib, Jess, and Alberto Bisin. 2005. "Modeling Internal Commitment Mecha- nisms and Self-Control: A Neuroeconomics Approach to Consumption-Saving Deci- sions." Games and Economic Behavior, 52(2): 460-92.

Bernheim, B. Douglas, and Antonio Rangel. 2004. "Addiction and Cue-Triggered Decision Pro- cesses." American Economic Review, 94(5): 1558-90.

Costa, Paul T., and Thomas A. Widiger. 1994. ed. Personality Disorders and the Five Factor Model of Personality. Washington, DC: American Psychological Association.

Fudenberg, Drew, and David K. Levine. 2006. "A Dual-Self Model of Impulse Control." Amer- ican Economic Review, 96(5): 1449-76.

Gul, Faruk, and Wolfgang Pesendorfer. 2001. "Temptation and Self-Control." Economet- rica, 69(6): 1403-35.

Laibson, David. 1997. "Golden Eggs and Hyper- bolic Discounting." Quarterly Journal of Economics, 112(2): 443-77.

Strotz, Robert. 1955-1956. "Myopia and Incon- sistency in Dynamic Utility Maximization." Review of Economic Studies, 23(3): 165-180.

Thaler, Richard H., and H. M. Shefrin. 1981. "An Economic Theory of Self-Control." Journal of Political Economy, 89(2): 392-406.

Article Contentsp. 966p. 967p. 968p. 969p. 970p. 971p. 972

Issue Table of ContentsThe American Economic Review, Vol. 97, No. 3 (Jun., 2007), pp. 543-1041, i-viiiFront MatterDistinguished Fellow [pp. 3-4]Macroeconomics for a Modern Economy [pp. 543-561]Relative Prices and Relative Prosperity [pp. 562-585]Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach [pp. 586-606]Generalizing the Taylor Principle [pp. 607-635]The Timing of Monetary Policy Shocks [pp. 636-663]Job Displacement Risk and the Cost of Business Cycles [pp. 664-686]Learning Your Earning: Are Labor Income Shocks Really Very Persistent? [pp. 687-712]Valuing New Goods in a Model with Complementarity: Online Newspapers [pp. 713-744]Estimating Risk Preferences from Deductible Choice [pp. 745-788]Estimating the Effects of Private School Vouchers in Multidistrict Economies [pp. 789-817]The Pluralism of Fairness Ideals: An Experimental Approach [pp. 818-827]Efficient Kidney Exchange: Coincidence of Wants in Markets with Compatibility-Based Preferences [pp. 828-851]Signaling Character in Electoral Competition [pp. 852-870]Harmonization and Side Payments in Political Cooperation [pp. 871-889]Meeting Strangers and Friends of Friends: How Random Are Social Networks? [pp. 890-915]Contracts and Technology Adoption [pp. 916-943]Leadership and Information [pp. 944-947]Social Interactions in High School: Lessons from an Earthquake [pp. 948-965]Measuring Self-Control Problems [pp. 966-972]Regulation, Capital, and the Evolution of Organizational Form in US Life Insurance [pp. 973-983]Sticky-Price Models and Durable Goods [pp. 984-998]Trust as a Signal of a Social Norm and the Hidden Costs of Incentive Schemes [pp. 999-1012]Tradeoffs from Integrating Diagnosis and Treatment in Markets for Health Care [pp. 1013-1020]ABCs (and Ds) of Understanding VARs [pp. 1021-1026]Matching and Price Competition: Comment [pp. 1027-1031]Effects of Environmental and Land Use Regulation in the Oil and Gas Industry Using the Wyoming Checkerboard as a Natural Experiment: Retraction [p. 1032-1032]Independent Auditors' Report [pp. 1033-1041]Back Matter

Documents

Measuring Self Control Problems